1 / 6

Configuration and Deployment of a scalable virtual machine cluster for molecular docking

Configuration and Deployment of a scalable virtual machine cluster for molecular docking. Nara Institute of Science and Technology, Nara Prefecture, Japan. Karen Rodriguez 7 /17/2013. Overview.

lula
Download Presentation

Configuration and Deployment of a scalable virtual machine cluster for molecular docking

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Configuration and Deployment of a scalable virtual machine cluster for molecular docking Nara Institute of Science and Technology, Nara Prefecture, Japan Karen Rodriguez 7/17/2013

  2. Overview • Virtual machines (VMs) have been observed to yield molecular docking results that are far more consistent than those obtained from a grid configuration. • Inhomogeneous results obtained from a grid are thought to be due to physical differences in the cluster’s components. This is eliminated by creating and networking cloned VMs. • This study’s objectives consist of constructing a clustered VM environment that is scalable according to job demand and which yields consistent dock results. • This system is to be tested, packaged, and deployed on the PRAGMA grid. This will provide sufficient computing resources to perform a full-scale docking project of large protein databases.

  3. Week 3: • Tested dock using test suite and compared results. We are close yet there is some discrepancy. • Resolved dock+MPI run issues such that jobs are sent to all machines and energy score results are consistent across the cluster (both for grid- and amber-based docking). • Tried compiling several input ligand files into one and running against a single receptor, results seem ok. • There are occasional firewall issues that cause MPI to not send jobs to all nodes • Experienced some problems with files changing back to default on their own.

  4. Future Plans • Find a permanent solution to firewall issues (firewall is turned off now, but need a long-term solution once it’s in the cluster). • Fix scripts to prevent defaulting dock input files whenever an error occurs. • Determine if energy score discrepancies are significant and if they can be easily dealt with (i.e., is it an offset?). If not, look into other types of compilers to use with dock. • Open MPI currently sends all jobs to all nodes. Although this has been useful as a test for result consistency on the cluster, we need to find a way to distribute jobs to nodes according to available processing power.

  5. This Week: Kyoto!

  6. Acknowledgments • Mentors • Dr. Jason Haga, UC San Diego Bioengineering • Dr. Kohei Ichikawa, Nara Institute of Science and Technology • UCSD PRIME Program • Teri Simas • Jim Galvin • Dr. Gabrielle Wienhausen • Dr. Peter Arzberger • Tricia Taylor • Funding • NAIST • Japanese Student Services Organization (JASSO) • PRIME • PRIME alumna Haley Hunter-Zinck • National Science Foundation

More Related